Research On Functional Module Detection From PPI Networks Based On Mutil Label Propagation And Artificial Bee Colony Optimization Mechanism

Protein interactions(Protein-Protein Interaction, PPI) network is a network of relationships biomolecular interaction between a living organism of all protein. Using computational methods to detect PPI network functional module in the post-genome field of bioinformatics is an active area of research. In recent years, new computational methods are emerging, but with the size of the network increases, how quickly and accurately identify PPI network functional modules is still an important research topic, attracting research interest to researchers.Multi-label propagation mechanism is transformed from the label propagation algorithm by simulating the actual network node belong to multiple categories of phenomena to find the optimal solution. Multi-label propagation algorithm does not require pre-defined objective function, it can within the time complexity of linear approximation to solve PPI network functional module detection. The algorithm converges very fast, but is easy to fall into local optimum. Colony optimization algorithm is a global optimization of intelligent algorithms, many studies have shown that the algorithm can find the global optimum in a finite number of iterations or approximate global optimal solution, but it is more of an individual solving iterative optimization process, so it takes more time. Based on the in-depth study of the multi-label propagation algorithm and aitificial bee colony optimization algorithm, this paper proposes two efficient PPI network functional module detection methods:(1) Inspired by multi-label propagation algorithm in complex network community discovery has a rapid and efficient ability, this paper presents a detecting functional module based on multi-label propagation mechanism in protein-protein interaction networks MLP-FMD. First of all, we combine the PPI network function information and structure information to initialize node label; then, according to the co expression of protein, the tag set is constructed, and the label is selected according to the threshold of the node to realize the update and dissemination of the label; finally, with tabs of the same protein nodes belonging to the same function module to obtain module test results. We apply the MLP-FMD algorithm to three data sets, and the results show that the MLP-FMD algorithm not only has a good time performance, but also has a certain competitive ability in the detection accuracy.(2) Aiming at the defection of multi-label propagation mechanism is easily trapped in the local optimum, we propose a functional module detection algorithm(LPABC-FMD) which combines the multi-label propagation mechanism with the artificial bee colony algorithm. Firstly, the algorithm uses the multi label propagation mechanism to initialize the population, which not only gets the initial population with high quality, but also reduces the scale of the problem solving; then, using the employed bees and the onlookers' neighborhood search for the optimal operation and the global search of the scouts int the artificial bee colony algorithm optimize the quality of the population; Finally, according to the automatic decoding mechanism to get the initial function module, and through the merger, filtering mechanism to get the final module. We have carried out experiments on LPABC-FMD algorithm and MLP-FMD algorithm on several data sets, and the results show that the LPABC-FMD algorithm can effectively jump out of local optimum. Compared with some classical algorithms, the LPABC-FMD algorithm has obvious advantages in detecting the quality of the functional modules.